What are conferences good for?

Mihai Pătraşcu wrote an elaborate post on the benefit of prestigious conferences. His arguments are simple enough:

  • Top conferences are communities created to increase the efficiency of the scientific community. You only have to monitor what happens at prestigious events instead of having to browse through everything that was published.
  • Yes, top conferences are biased toward certain topics, and that’s a good thing.
  • It most efficient to assess a researcher by the number of papers he published at top conferences.
  • Journals have a worse review process than conferences.

His points are good and well made. But I oppose his ideas nevertheless.

In practice, I do not rely on lists of accepted papers to follow what happens in my field. Tools like Google Scholar allow me to quickly browse all papers matching some keywords. It has proven to be a sufficiently good recommender system. I try hard to avoid having a field or belonging to a fixed community. I try hard to think differently.

Self-reinforcing biases are bad and must be fought against. They are killing Physics and have supported 40 years of ill-fated research in classical AI. I feel we have “too much of the same” and I wish to encourage my fellow researchers to be more daring. I want to read researchers who are challenging preconceived ideas, not researchers carefully copying the trends.

I also think it is deeply wrong to think that you can assess a researcher without reading his work. Parnas said it best:

When serving on recruiting, promotion, or grant-award committees, read the candidate’s papers and evaluate the contents carefully. Insist that others do the same.

In short, I do not believe you should try to belong to the community defined by top conferences. It is akin to trying to fit in with the cool kids in high school. I am sorry, but I was and will remain a misfit for this reason alone:

The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man. (George Bernard Shaw)

Finally, some journals have a terrible review process, but I think that Pătraşcu is just misguided to think that “journal acceptance is determined by one dude (the editor), who’s not even anonymous, based on some inconclusive statement from another dude.”

Note. I have a cool tenured job. Maybe graduate students should follow Pătraşcu’s advice. Myself, I do not plan to apply for another academic job in the near future.

7 thoughts on “What are conferences good for?”

  1. Hey, comment 1 is from me. I didn’t mean to be Anonymous. It’s just that my browser usually fills in the form for me, but I’m not using my regular browser.

  2. I fail to see the difference between conference and journal reviewing, except that conference papers are shorter and therfore lack detail. Still “dude tells dude his opinion”.

    Ultimately, yes, but my experience has been that journal editors require a conclusive peer review before accepting a paper.

    Now specialized conferences are great for creating intensive atmospheres, especially if the encourage work-in-progress submissions of high quality. This likely involves a different reviewing process than many prestigious meetings.

    My experience has been that prestigious conferences tend to proudly favor a small set of topics and directions as “interesting”. Eventually, all papers end up having the same biases.

    What machine learning and democracy taught us is that diversity matters a whole lot. Packing a room full of people who think exactly alike is a recipe for disaster. That’s not how science moves forward.

    If it we worked that way, we would still have alchemy conferences. It takes a chemist to tell the alchemists they are wrong. You want to invite the people who disagree with you at your party.

  3. Daniel, I agree completely with your arguments. On the stay-at-home side, you should add concern for the impact of air travel on climate change. On the go-to-a-prestigious-conference side, there is the appeal of travel to exotic locations. 🙂

  4. I fail to see the difference between conference and journal reviewing, except that conference papers are shorter and therfore lack detail. Still “dude tells dude his opinion”.

    Now specialized conferences are great for creating intensive atmospheres, especially if the encourage work-in-progress submissions of high quality. This likely involves a different reviewing process than many prestigious meetings.

  5. Although I’ve not been to any of the prestigious conferences in my field, I do keep track of the papers. In this sense, my experience mirrors yours: the papers become homogenous over time. In my work, to borrow a phrase from one of Mike Myers’ characters on Saturday Night Live, “If it’s not type theory, it’s crap!”

    It may not be good for my academic career to eschew these prestigious conferences, but I have my doubts I would be able to get a paper into them since I am essentially arguing the opposite — and I don’t have any previous clout.

    A lot of the ideas I’ve been putting together have come from revisiting old ideas and Google Scholar searches, not looking through what gets published in the well-regarded conferences and journals.

    I guess I share some of the rebel spirit. 🙂

  6. A lot of the ideas I’ve been putting together have come from revisiting old ideas and Google Scholar searches, not looking through what gets published in the well-regarded conferences and journals.

    I guess I share some of the rebel spirit. 🙂

    That’s what research is supposed to be. And once people read your papers, they should be able to appreciate your results, whether or not they appeared in one of the trendy conferences.

    Of course, if they just look for the name of a prestigious conference, you might get in trouble… but anyone who reviews your work without reading your papers is a fraud. He may be a fraud with power, but he is, nevertheless a fraud.

    There is a theory going around that reading someone’s papers is a lot of work. To read them carefully is hard, yes… but to browse and read random sections is not that hard. And you quickly get a feel for what the researcher is after… whether he is just following trends or building up an original research program. Of course, this assumes that you are interested and know the area a bit, but experienced researchers get to know several areas (unless they belong to one of these focused communities).

Leave a Reply

Your email address will not be published. Required fields are marked *